IEEE TRANSACTIONS ON SMART GRID, VOL. 5, NO. 4, JULY

2y ago
10 Views
3 Downloads
1.90 MB
9 Pages
Last View : 18d ago
Last Download : 2m ago
Upload by : Ronnie Bonney
Transcription

IEEE TRANSACTIONS ON SMART GRID, VOL. 5, NO. 4, JULY 20142075A Distributed Demand Response Control StrategyUsing Lyapunov OptimizationLei Zheng, Student Member, IEEE, and Lin Cai, Senior Member, IEEEAbstract—Motivated by the potential ability of heating ventilation and air-conditioning (HVAC) systems in demand response(DR), we propose a distributed DR control strategy to dispatchthe HVAC loads considering the current aggregated power supply(including the intermittent renewable power supply). The controlobjective is to reduce the variation of nonrenewable power demand without affecting the user-perceived quality of experience.To solve the problem, first, a queueing model is built for thethermal dynamics of the HVAC unit based on the equivalentthermal parameters (ETP) model. Second, optimization problemsare formulated. Based on an extended Lyapunov optimization approach, a control algorithm is proposed to approximately solve theproblems. Third, a DR control strategy with a low communicationrequirement is proposed to implement the control algorithm in adistributed way. Finally, practical data sets are used to evaluateand demonstrate the effectiveness and efficiency of the proposedcontrol algorithm.Index Terms—Demand response (DR), heating ventilation andair-conditioning (HVAC), Lyapunov optimization, power variation, renewable power integration, smart grid, thermal dynamicqueue.I. INTRODUCTIONDEMAND response (DR) is anticipated to be a critical application in smart grid. Aided by the advanced meteringinfrastructure (AMI), the power usage of different appliances inthe customer premises can be adjusted either directly, such asoperational parameters/states changing requested by grid operators; or indirectly, such as real-time pricing. By smoothing outthe system power demand over time, DR is capable of providingpeak shaving, load shifting and ancillary services to maintainthe system reliability and stability.On the power supply side, a growing number of renewableenergy sources are introduced into the power grid. The renewable energy can reduce congestion in the grid and decrease theneed for new generation or transmission capacity. However, theintermittent nature of renewable energy brings new challenges,which can be inimical to the power grid stability, and requiresextra energy storage or local generation to balance the generated power with the demand. Thus, the potential positive environmental and economic benefits may be offset by these newproblems and costs [1].Manuscript received May 01, 2013; revised December 18, 2013; acceptedMarch 11, 2014. Date of current version June 18, 2014. Paper no. TSG-003432013.The authors are with the Department of Electrical and Computer Engineering,University of Victoria, Victoria, BC V8W 3P6, Canada (e-mail: zhengl@ece.uvic.ca; cai@ece.uvic.ca).Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.Digital Object Identifier 10.1109/TSG.2014.2313347On the customers side, the customer power demand cantypically be divided into three categories, inelastic load andtwo types (Type-I and Type-II) of elastic load. The inelasticload must be satisfied immediately when needed, e.g., lighting.Hence, the inelastic load is not suitable for DR. The Type-Ielastic load includes the power demand of the devices whoseoperation can be delayed but not interrupted, such as washers.For DR, this type of demand is mostly interested in providingpeak shaving and load shifting services. The Type-II elasticload denotes the most flexible power demand, such as heatingventilation and air-conditioning (HVAC) systems. Considering the thermal capacity of the building, which introducescorrelation of the temperature across time and is similar to aqueueing system, the control of HVAC units can align wellwith the needs to smooth the energy demand variation in thetime scale of minute-level. The potential of HVAC devices forload balancing/regulation service has been evaluated in [2].In the literature, there have been many works on how to useDR to shave demand peaks or to shift the peak [3]–[8]. Whileboth of the power peak and the power variation are importantto the stability of the power systems, the later one fluctuates ina much smaller time scale (minute-level) with a relatively lowamplitude comparing to the demand peak. In this paper, motivated by the HVAC units’ potential in demand response service,our focus is to explore how to utilize in-house HVAC units to reduce the power demand variation, which has not attract enoughattention previously. By smoothing the energy demanding inthe minute-level, the total cost for the power generation canbe reduced, as we can reduce the needs for online regulationservices [9], [10].The main contributions in this paper are fourfold. First, webuild a queueing model for the thermal dynamics of HVACunits, a representative source of the Type-II elastic load. Withsuch a queueing model, the controlled room temperature is similar to the power in a battery, which is increased (filled) whenthe HVAC unit is on (when the battery is charged) and is decreased (emptied) when the HVAC unit is off (when the batteryis discharged). Second, optimization problems are formulatedto minimize the average variation of the nonrenewable powerdemand by controlling the on/off states of HVAC units. By extending the Lyapunov optimization techniques in [11], we canjointly optimize the objective value and guarantee the room temperatures staying in customers’ desired regions. Third, to further reduce the communication cost and complexity, we propose a suboptimal control algorithm and a strategy to implementthe algorithm in a distributed way. One more merit of the control algorithm is that it can be tuned to effectively reduce theaverage variation of the nonrenewable power demand without1949-3053 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See http://www.ieee.org/publications standards/publications/rights/index.html for more information.

2076IEEE TRANSACTIONS ON SMART GRID, VOL. 5, NO. 4, JULY 2014significantly increasing the frequency of HVAC units’ on/offswitching. At last, using practical data sets, simulation resultsdemonstrate that our control algorithm can effectively reducethe variation of nonrenewable power demand and guarantee thecustomers’ comfortable experiences.The rest of this paper is organized as follows. In Section II, wepresent a summary of related work on DR control. Section IIIdescribes our system model. In Section IV, we present ourqueueing model for the HVAC units’ thermal dynamics. InSection V, optimization problems are formulated to minimizethe average variation of nonrenewable power demand and asuboptimal control algorithm is proposed. In Section VI, acontrol strategy is proposed to implement our control algorithmin a distributed way. Section VII presents the numerical resultsfollowed by the concluding remarks and future research issuesin Section VIII.II. RELATED WORKWith customers’ participation, DR enables more options tobalance the power supply and demand. To attract customers’participation, one important strategy is to shape the powerdemand through time-dependent pricing (TDP) [4]. By monitoring the electricity consumptions and providing customersthe real-time price information through the smart grid infrastructure, the power operators can manipulate the electricityprice. Thus, customers’ electricity usage may be restrained atthe high power price period and stimulated at the low powerprice period. Note that the controllable power demand can beany of the three types discussed above, when the electricityusages are controlled by customers according to the TDP. In[5], game theory has been applied to reduce the peak load ofthe grid considering the consumers’ reactions to the electricityprices, the subsequent changes in the demand pattern of thetarget day, and the resulting effect on observed prices. The benefits of using TDP are illustrated in [12], which used real datato illustrate that shifting usage physically can reduce the riskof overloading. However, one limitation of the pricing basedDR strategy is that it relies on the assumption of reasonablecustomers’ reaction to the electricity price, which may notalways be true in a real environment. Besides, it also dependson the price prediction to achieve such benefit, which is also achallenging issue.Another type of DR is to control the demand-side loaddirectly by utilities or system operators. Considering thecustomers’ requirements on comfortable experiences, most researches focus on utilizing the Type-I elastic load. Dividing theloads into real-time loads (inelastic load) and controllable loads(Type-I elastic load), [6] proposed an approach that attemptsto produce a uniform load demand over time by schedulingthe power usage lower than a preset target. In [3], a stochasticmodel was developed and two online demand scheduling policies were introduced to minimize the long-term average powergrid operational cost. In the first one, the controller serves anew demand request immediately or postpones it to the end ofits deadline, depending on the current power consumption. Inthe second one, a new power demand is activated immediatelyif power consumption is lower than a threshold; otherwise itis queued. A queued demand is activated when its deadlineexpires or when consumption drops below the threshold.Fig. 1. Demand response in smart grid.Comparing to the Type-I elastic load, Type-II elastic loadcan be more attractive for DR as the power can flow in twodirections, not only help reduce the load demand in a period,but also compensate the power shortage by being “discharged”without requiring energy storage devices, i.e., functioning asbattery. In the literature, there are also several strategies takingthe advantage of the Type-II elastic load for the direct DR. [13]presented a strategy for that using water heaters as regulationresources. In [14], a direct control strategy was proposed tomanage the large population of HVAC units using the systemidentification approach. A centralized optimal control algorithmwith comfortable room temperature consideration was proposedin [15] by controlling the operational set-point of HVAC units.However, the control algorithm in [15] relies on the populationinformation of the room temperature, which makes it vulnerable if the data packet is lost due to communication impairments [16]. Nevertheless, utilizing the power-storage feature ofType-II elastic load is a promising approach to provide ancillaryservice in smart grid, which motivates our work in this paper.III. SYSTEM MODELIn this paper, we aim to reduce the variation of nonrenewablepower demand, which is typically supplied by traditional powerplants, by controlling the “ON/OFF” states of HVAC units.Fig. 1 shows a typical demand response scenario in smartgrid. In the service community, there aredistributed residential houses, assuming each of the house is equipped with anHVAC unit and a smart meter. A control center connects thecustomers to the renewable and nonrenewable power sourcesthrough communication networks, and directly controls customers’ HVAC working states through communicating with thesmart meters.To satisfy customers’ requirements on comfortable living environments, a comfortable temperature region is set for each residential house. Let be the room temperature in the th house) anddenote the comfortable tem(perature region, such that. In each house, isdetermined by the environment temperature ( ), the previousroom temperature and the working state of the HVAC unit.Letbe the state of an HVAC unit in slot ,be theequivalent heat capacity (C),be the equivalent thermal) of the residential house, andbe theresistance ( Cequivalent heat rate ( ) of the HVAC unit. According tothe ETP model in [2], the room temperature ( C) evolves asfollows:ifo.w.,(1)

ZHENG AND CAI: DISTRIBUTED DR CONTROL STRATEGY USING LYAPUNOV OPTIMIZATIONwhere. An HVAC unit only switches itsworking state when the room temperature reaches one of theregion bounds, i.e., switching from “ON” () to “OFF”() if, switching from “OFF” to “ON” if, and keeping its working state when.In this paper, it is assumed that the environment temperatureis the same for theresidential houses but changes overtime as a random variable and. For the HVAC,,,thermal dynamics related parameters, including, and, they can be different for different customers andtheir houses.To implement the DR control, time is divided into time slotswith slot duration. Instead of letting HVAC units work automatically, control decisions are made in each time slot to designate the working states of HVAC units.On the customer side, at each time slot, we divide the load de)mand into two parts: HVAC load and non-HVAC load (that includes the inelastic load and Type-I elastic load. On thesupply side, the power is supplied by two kinds of sources, thetraditional power grid and the renewable power sources, e.g.,wind power, denoted asand, respectively. We assume that there exists a peak power supplyand a peakload demand, so thatand.In time slot , the power supplies should be equal to thecustomers’ total loads, which requires that2077Fig. 2. Validation of the queueing model of HVAC thermal dynamics (W,C W,C,C).Comparing (3) and (4), we can derive that(5)(6)Whentoin, the room temperature will increase fromslots. Suppose(7)(2)is the power consumed by an HVAC unit if it is turnedwhereon.However, as the renewable power () is time-varying andnoncontrollable, the power supplied by the traditional powergrid () has to vary timely, which causes great challengesto the power generation and the grid stability. To minimize thevariation of the demand on the traditional power supply ()caused by these time-varying power supplies, we propose an approach to balance the load demand and power supply throughtuning the load demand by directly controlling the HVAC units’“ON/OFF” states but not disturbing customers’ comfortable experiences. We assume that customers have the incentive to participate in the direct DR control, as they will be compensated bythe power company accordingly.IV. QUEUEING MODEL OF HVAC THERMAL DYNAMICSIn this work, considering that the thermal capacity of thebuilding introduces correlation of the temperature across time,which is similar to a queueing system, we remodel the HVACthermal dynamics in (1) using a queueing model. Letbe the temperature loss in each time slot, andbe theroom temperature increased by the HVAC unit if it is on. Giventhe environment temperature, when, the roomtointemperature will decrease fromslots. Suppose(3)and according to (1), we have(4)Similar to the case when, we have(8)(9)By (1) and (3)–(9), we derive the queueing model of theHVAC thermal dynamics as follows:(10)To evaluate the accuracy of the queueing model, simulationhas been run to compare the room temperature dynamics of anHVAC unit in 250 minutes based on the proposed model and theone in [2]. According to the results shown in Fig. 2, the proposedmodel matches the one in [2] quite well.V. OPTIMAL DEMAND RESPONSE CONTROLBased on the queueing model in Section IV, we study howto utilize HVAC units to reduce the average variation of nonrenewable power demand by using the mean square successivedifference ofas our optimization objective.A. Problem Formulationandcan be meaIt is assumed that the currentsured or estimated [17]. Let.represents the difference of the nonrenewable power demand in two successive time slot. Thus, ouroptimization problem is formulated as P1.

2078IEEE TRANSACTIONS ON SMART GRID, VOL. 5, NO. 4, JULY 20141) Problem I (P1):3) Problem III (P3):(11)(19)s.t.(12)(13)where (12) stands for each customer-desired room temperaturerequirement and (13) says that each HVAC unit can either be onor off.The problem above is challenging mainly due to thetime-coupling property brought by the first constraint. Previous methods handling similar problems are usually basedon dynamic programming, requiring detailed knowledge ofstatistics ofand, and are vulnerable to the curse ofdimensionality problem [18]. Moreover, these statistics may beunknown or difficult to obtain in practice.Another way to solve P1 is to study its relaxed form by usingthe bounded average room temperature instead of constraint(12); thus:2) Problem II (P2, Rhe Relaxed P1):s.t.(20)In each time slot,,are updated according to (2)and (10), respectively, and the optimal control decision ofcan be found by solving P3.Theorem 1: The solution to P3 is always feasible to P1.Before we prove Theorem 1, we first present the solution toP3. Let. We first sort the HVAC unitsin an ascending order. Let be the indicatoraccording toof the order of th HVAC unit anddenote the HVAC unitbe the th one in the sorted sequence. We obtain the optimalsolution to P3 by finding the th HVAC unit satisfying (21)and (22) as follows:(21)the same ass.t.(14)(15)whereis the average room temperature.The solution to P2 is easy to be characterized based on theframework of Lyapunov optimization [11]. However, the solution for the relaxed problem may not be feasible for the originalproblem.In this work, we introduce auxiliary parametersfor each HVAC unit and virtual temperaturequeues,, as a shift of. We have(16)(22)denotes the increment of (11) when the th HVACwhereunit is turned on.Thus, the control decisions for all HVAC units are(23)withTheorem 1.Proof: Assuming that(24) when. In the following, we prove, with (17), we obtain, and (25) when, respectively:where(24)(25)(17)With (10) and (16), we obtainvirtual queues as(18)and then we can reformulate P1 as a quadratic optimizationproblem based on the Lynapnuv optimization [11]. After somemanipulations (Please refer to the Appendix.), we obtain:is monoAs the functiontonically increasing, the control decision iswhenandwhen, such that no room temperaturewill be out of the customer desired region, which satisfies theconstraint in (12). Note that, for the HVAC units with room temperature betweenand,they can be either turned on or off.Theorem 2: Ifis i.i.d. over slots, then the expectedunder our algorithm is within a boundof the optimalvalue(26)

ZHENG AND CAI: DISTRIBUTED DR CONTROL STRATEGY USING LYAPUNOV OPTIMIZATIONwhere(27)Proof: First, P1 is relaxed to be P2 with the mean ratestable constraint only. Letbe the optimal value ofofP1 andbe the optimal value ofin P2. As all thesolutions to P1 will be feasible to the relaxed problem and thereare looser constraints in P2 than in P1,.Second, with (16), we can formulate a new optimizationproblem P4 (as shown in the Appendix.) with the same objective as that in P1, but with constraints on mean rate stable ofthose virtual queues only. Letdenote the optimal value ofthe objective in P4. To solve P4, we actually solve P3. It can beproved using Lyapunov theory [11] that there exist a constant,.Third, comparing P2 and P4, they are of the same form exceptthe different meaning of the queues. Thus,. We derive that the achievable objective is bounded by.B. DR Control AlgorithmOur control algorithm using the extended Lyapunov optimization above is summarized in Algorithm 1.Algorithm 1 The DR control algorithm1: Setsatisfying (17)2: In each time slot3: collect,of each customer’house,andin the last slot4: update the values of5: sort the virtual queue according to6: findaccording to (21)-(22)if, otherwise,7: For each HVAC unit,Note that, when different combinations of the auxiliary parameterfor thequeues are used, the control decisionsmade may be different, so as the room temperatures, which willbe demonstrated in Section VII. However, whensatisfies(17), the result of the control objective will be almost the same.VI. DISTRIBUTED DEMAND RESPONSE CONTROL STRATEGYTo implement Algorithm 1, one strategy is to use a centralized control by collecting all required information to the controlcenter and then delivering theto each customer every slotafter decision-making. However, such a strategy is not efficientas these information (line 3 in Algorithm 1) has to be reportedto the control center frequently, which introduces large communication cost and time delay for decision-making. In addition,such a strategy is vulnerable to security and privacy problems asboth customers’ private information and control decisions maybe intercepted during the communication process.Observing the solution to P3 in (21)–(23), it is found that theHVAC unit’s state in a time slot depends on the sortedand corresponding order . While,,,andcan be known by each customer, the only informa-2079tion required for customers to make their own decisions on theHVAC units’ states is the value ofand . Thus, in thispaper, we propose a distributed DR control strategy, as shownin Strategy 1.In a control slot, the control center distributes a summary ofpower consumptionand the virtual queue sequence ( )to all customers. Accordingly, each customer makes the decision independently by checking whetheris positive or notaccording to its own virtual queue sequence . On the otherhand, the control center can also predict customers’ decisions by(21)–(23), the room temperatures by (10), and the virtual queuelength by (16).With such a distributed implementation, the benefits of sucha strategy are three-folds. First, the communication cost can bereduced with fewer customer reports than that in centralizedcontrol; second, the control can be more reliable as there is nocontrol-error due to the communication error in delivering thecontrol decisions from the control center to customers; third, thesystem can be more secure with a lower frequency for customersto report their private information and no control decision is delivered over communication networks.Strategy 1 The distributed DR control strategy1: In each time slot ,2: the control center calculates the summary of powerconsumption () and the virtual queue sequence ( )and to customers3: the control center delivers4: each customer calculate its own5: ifthen6: the customer turns on its HVAC unit during the slot7: else8: the customer turns off its HVAC unit during the slot9: end ifNote that it is possible to have an inaccurate prediction ofthe household’s room temperature as we use a queueing modelfor approximation. However, the difference will be quite smallwithin a slot when the control slot is short, e.g., one minuteas shown in Fig. 2. To avoid the inaccurate prediction to accumulate to the point of causing a negative effect, one approachis to increase the frequency of customers reporting their roomtemperatures so that the control center can limit the error. Howto quantify the inaccurate prediction and design an optimal reporting interval to balance the control benefit and communication cost [19] is left for future study.VII. PERFORMANCE EVALUATIONA. Simulation SettingsWe evaluate the proposed DR control algorithm in Section Vin a community with 2000 residential houses and a 1-MW windturbine providing renewable energy using practical data. For thepower supply, the renewable power data are generated with atypical turbine power-curve using the wind speed data duringApr. 10–12, 2012 taken from Canada climate website, http://climate.weather.gc.ca. On the customer side, we used typical res-

2080IEEE TRANSACTIONS ON SMART GRID, VOL. 5, NO. 4, JULY 2014Fig. 3. Environment data. (a) Wind turbine power-curve. (b) 24-hour wind speed. (c) 24-hour environment temperature.TABLE ISIMULATION PARAMETERSFig. 5. Mean variation of the nonrenewable power demand (Fig. 4. Nonrenewable power demand.idential house non-HVAC loads [20]. While the data from [20]are the discrete average load per 5 minutes, we interpolatedit into per minute loads with Gaussian fluctuation, which is10% of the load on average. The environment temperature data,which are required for the proposed control algorithm, are alsofrom Canada climate website. Fig. 3 shows a) the wind turbinepower-curve (cut-in speed: 3 m/s, cut-out speed: 20 m/s, ratedpower output: 1 MW), b) the 24-hour wind speed and c) the24-hour environment temperatures used in the simulation., as one minute,For DR control, we set the control slot,which is short enough that the customers’ load demand and therenewable energy supply are assumed static. Table I presentsthe HVAC units related parameters. For the HVAC thermal dy, andnamics related parameters, including, we assume they are uniformly distributed in the rangeshown in Table I. For the HVAC unit power load, it is assumedto be 600 W.B. Simulation ResultsIn this section, the proposed DR control algorithm is firstcompared with two other schemes: one without DR control andone using the algorithm in [15], in which the HVAC units arecontrolled by adjusting the customers’ set-point of the room.temperatures within).1) Control Effectiveness: Fig. 41 shows the power demandfor the nonrenewable energy. It is observed that the power demand without DR control in Fig. 4 fluctuates seriously due to theload variation and the intermittent renewable power supply. Bycontrast, both the proposed algorithm and that in [15] effectivelyreduce the fluctuations, which brings down the risk of poweroutage and reduces the need for activating high cost supplementary power generation sources for load balancing/regulation.Fig. 5 presents the average nonrenewable power demand differences in consecutive slots. Although the variation of the nonrenewable power supply seems larger using the proposed schemethan that in [15] occasionally in Fig. 4, the overall performance ofthe control algorithm outperforms that in [15] substantially witha 19% gain. When compared to that without DR control, a 32%gain is achieved by the proposed algorithm in the simulation.This is because the proposed scheme directly control the HVACunits’ states instead of attempting to affect the states through anyintermediate variable, i.e., the room temperature set-point, whichenables a finer granularity to tune the loads.2) Cost of the Control Algorithm: While the HVAC units arecontrolled to reduce the variability of power production, one potential cost is the increase of the frequency of the HVAC on/offswitching. In Fig. 6, such impact is evaluated by the PMF ofthe number of HVAC on/off cycles per hour. It is found that thenumber of on/off cycles increased from about 0–3 cycles perhour without DR control to about 1–5 cycles per hour using theone in [15], and to about 5–20 cycles per hour using the proposed control algorithm.For the control algorithm in [15], as the set-point is controlled, only the HVAC units, with which the room temperaturesare close to or , will toggle their states and others will keeptheir states. Thus, its impact on the frequency of HVAC on/off1In the figure, we enlarge the result between 964–1002 minutes. Please referto [21] for the enlarged version of the figure.

ZHENG AND CAI: DISTRIBUTED DR CONTROL STRATEGY USING LYAPUNOV OPTIMIZATION2081Fig. 8. Mean variation of the nonrenewable power demand with the adaptive(,,,).Fig. 6. PMF of the HVAC on/off cycles per hour ().Fig. 7. PMF of the HVAC on/off cycles per hour with the adaptive,,,).(switching is limited. Comparing to [15], the proposed algorithmdirectly controls the HVAC units’ states. In every slot, both ofthe previous on and off HVAC units may change their states,which causes more frequent HVAC on/off switching.To avoid overusing the HVAC units, we can try to keep theHVAC units’ states as much as possible. To do so, instead ofusing the constantin (16), we can use a time varying,which is related to the HVAC unit’ previous state, andWhen, it is the same as using . When, if anHVAC unit is previously turned on, it will get a large shift foritsto have a low order in ; otherwise, it is likely to gaina high order. In this way, HVAC units will be more likely tokeep their states to avoid frequent on/off switching. As shownin Fig. 7, with a smaller , there is high probability to have theHVAC on/off switch less than 5 cycles per hour.On the other hand, with the adaptive, the control effectiveness may also be affected. Fig. 8 shows the power variationwith different values of . As it is shown, the mean variationof nonrenewable power demand increases as the value of decreases. Thus, an adaptivemay help balance the cost andthe effectiveness of the proposed control algorithm. Also, it ispossible to devise an incentive mechanism to encourage userstolerating a larger value of to provide more DR, which is leftfor future research.Fig. 9. Residential house room temperature sample (C).C,Note that, with the adaptive,may changes with alarger fluctuation but still bounded (shown later in Fig. 9). Theproof for this can be found in [21].3) Impact on Customers’ Comfort Requirements: At last, weevaluate the impact of the DR control algorithm on the customer’s comfort requirements by showing the room temperature of a sample residential house in Fig. 9. As it is shown,the proposed algorithm guarantees that the desired temperaturerequirements. However, the algorithm in [15] violates the desired room temperature setting sometimes. It is also found thatthe proposed scheme can provide more comfortable experiencefor a customer as the room temperatures is more stable with asmaller gap between the desired region. Meanwhile, when theadaptiveused, the room temperature controlled by theproposed DR control algorithm fluctuates in a larger range witha small .VIII. CONCLUSION AND FUTURE WORKIn this paper, we have studied the DR control using HVACunits. A DR control algorithm based on the Lyapunov optimization has been proposed. Simulations with practical data setshave showed that the proposed control algorithm is effectivein reducing the variation of the nonrenewable power demandand guaranteeing customers’ comfortable experiences. Besides,a distributed strategy to implement the control algorithm hasbeen proposed, which has fewer communication cost and moresecure. Moreover, simulation results demonstrates that the proposed algorithm

IEEE TRANSACTIONS ON SMART GRID, VOL. 5, NO. 4, JULY 2014 2075 A Distributed Demand Response Control Strategy Using Lyapunov Optimization Lei Zheng, Student Member, IEEE,andLinCai, Senior Member, IEEE Abstract—Motivated by the potential ability of heating venti- lati

Related Documents:

IEEE 3 Park Avenue New York, NY 10016-5997 USA 28 December 2012 IEEE Power and Energy Society IEEE Std 81 -2012 (Revision of IEEE Std 81-1983) Authorized licensed use limited to: Australian National University. Downloaded on July 27,2018 at 14:57:43 UTC from IEEE Xplore. Restrictions apply.File Size: 2MBPage Count: 86Explore furtherIEEE 81-2012 - IEEE Guide for Measuring Earth Resistivity .standards.ieee.org81-2012 - IEEE Guide for Measuring Earth Resistivity .ieeexplore.ieee.orgAn Overview Of The IEEE Standard 81 Fall-Of-Potential .www.agiusa.com(PDF) IEEE Std 80-2000 IEEE Guide for Safety in AC .www.academia.eduTesting and Evaluation of Grounding . - IEEE Web Hostingwww.ewh.ieee.orgRecommended to you b

Signal Processing, IEEE Transactions on IEEE Trans. Signal Process. IEEE Trans. Acoust., Speech, Signal Process.*(1975-1990) IEEE Trans. Audio Electroacoust.* (until 1974) Smart Grid, IEEE Transactions on IEEE Trans. Smart Grid Software Engineering, IEEE Transactions on IEEE Trans. Softw. Eng.

Citation information: DOI 10.1109/TSG.2017.2664043, IEEE Transactions on Smart Grid IEEE TRANSACTIONS ON SMART GRID, VOL. XX, NO. XX, MONTH YEAR 1 Spoofing-Jamming Attack Strategy Using Optimal Power Distributions in Wireless Smart Grid Networks Keke Gai, Student Member, IEEE, Meikang Qiu, Me

IEEE TRANSACTIONS ON SMART GRID 1 Preventing Occupancy Detection From Smart Meters Dong Chen, Student Member, IEEE, Sandeep Kalra, Student Member, IEEE, David Irwin, Member, IEEE, Prashant Shenoy, Fellow, IEEE, and Jeannie Albrecht, Member, IEEE Abstract—Utilities

686 IEEE TRANSACTIONS ON SMART GRID, VOL. 2, NO. 4, DECEMBER 2011 Multicast Authentication in the Smart Grid With One-Time Signature Qinghua Li, Student Member, IEEE, and Guohong Cao, Fellow, IEEE Abstract—Multicast has been envisioned to be useful in many smart grid

systems, microgrids, smart grid, storage. Abstract The IEEE American National Standards smart grid publications and standards development projects IEEE 2030, which addresses smart grid interoperability , and IEEE 1547 TM, whichaddresses distributed resources interconnection with the gridhave made substantial , progress since 2009 [1]. The IEEE 2030

446 IEEE TRANSACTIONS ON SMART GRID, VOL. 4, NO. 1, MARCH 2013 An Information-Theoretic Approach to PMU Placement in Electric Power Systems Qiao Li, Student Member, IEEE,TaoCui, Student Member, IEEE,YangWeng, Student Member, IEEE, Rohit Negi, Member, IEEE, Franz Franchetti, Member, IEEE, and Marija D. Ilić, Fellow, IE

IEEE TRANSACTIONS ON SMART GRID, VOL. 7, NO. 3, MAY 2016 1683 Provision of Regulation Service by Smart Buildings Enes Bilgin, Member, IEEE, Michael C. Caramanis, Senior Member, IEEE, Ioannis Ch. Paschalidis, Fellow, IEEE, and Christos G. Cassandras, Fellow, IEEE Abstract—Regulation service (RS) reserves, a critical type of bi-direc